Nvidia’s ‘Sovereign’ AI Could Win a Prize for Irony

Nvidia Corp. CEO Jensen Huang has been actively promoting his vision of 'sovereign' artificial intelligence during his recent travels across Europe. Dressed in his trademark leather jacket, Huang has shared the stage with notable figures such as Keir Starmer and Emmanuel Macron, advocating for the establishment of new data centers that would provide essential computing power within national borders, rather than relying on dominant tech firms from abroad. This endeavor resonates with European leaders seeking innovative solutions to enhance productivity and mitigate the risks of falling behind in a tech landscape heavily influenced by the U.S. and China. Nvidia's recent partnerships, including one with the French AI startup Mistral to create a cloud platform utilizing 18,000 Nvidia Blackwell chips, and another aimed at developing an industrial cloud for European manufacturing with 10,000 Blackwell chips, underscore this ambition. Huang is particularly optimistic about the potential for computing capacity in Europe to increase tenfold within the next two years, stating, 'It’s coming.' However, the irony of this 'sovereignty' claim cannot be overlooked. From Nvidia's perspective, the company is positioning itself as a key geopolitical player, engaging directly with heads of state like Macron to facilitate AI adoption. This strategy is particularly advantageous for Nvidia, especially in light of a broader Sino-American trade conflict that has resulted in significant sales losses in China due to export restrictions. Bloomberg Intelligence recently projected that investments in sovereign AI could generate an additional $10 to $15 billion in annual revenue for Nvidia in a world moving towards de-globalization. Yet, while leaders like Macron advocate for tech autonomy to reassure their constituents, the reality remains that Europe still heavily relies on Nvidia, a U.S.-based company that commands an 80% market share in the AI chip market. Despite some advancements, Europe’s current AI capabilities are limited. With Mistral raising €1 billion ($1.2 billion) in funding, the contrast with the substantial backing received by competitors like OpenAI is stark. Currently, Europe accounts for only 4.8% of the global AI computing power, as reported by Epoch AI. Historically, attempts by France and Germany to create a sovereign cloud to safeguard user data from U.S. and Chinese influence have fallen short, with U.S. tech companies still dominating approximately 70% of cloud services in Europe. Instances like Danish municipalities moving away from Microsoft stand in contrast to the German military's cloud agreement with Google, highlighting the ongoing reliance on American technology. Even European alternatives to Google often depend on Microsoft’s Bing, leading to failures when Bing experienced outages. Ironically, Microsoft is now marketing 'sovereign cloud' services to Europe, seemingly without recognizing the underlying contradictions. The broader question remains whether Huang's vision of AI as a foundational technology akin to electricity or the steam engine will materialize. At a recent AI summit, a panel I moderated likened AI sovereignty to a national airline, where the operations matter more than the origin of the components. Yet, failing to address dependency issues—like reliance on Russian gas or Chinese exports—could pose significant risks for AI development as well. For Europe to genuinely achieve AI sovereignty, it must do more than repackage existing technology. It has the potential through its talent pool, skilled workforce, and companies like ASML Holding NV. However, the continent currently lacks the robust research funding, financing, and end-user demand that have enabled U.S. startups to outpace their European counterparts significantly. Addressing these gaps will require time and strategic investment in chip independence to cultivate viable alternatives and mitigate risk. Without such efforts, Europe risks falling into a state of digital colonialism as global competitors tighten their grip on technology. As Huang continues to champion this vision, the need for true European autonomy in AI has never been more pressing.